Transcript + (+1) - Winthrop University

Attitude Measurement
What is an Attitude
• A mental state used by individuals to structure
the way they perceive the environment and guide
the way they respond to it
• Essence of the ‘human change agent’ –
influencing attitudes can influence how you
behave
• Great diagnostic / explanatory value – why
consumers buy / don’t buy
• Overwhelming amount of primary research in
marketing deals in measuring attitudes
Formation of attitudes (MAAM)
• Belief about a brand = Attribute x strength of its
association with brand
• Importance of attribute moderates belief strength
• Sum of moderated beliefs = attitude to brand
• Interpret the figure according to the direction of
the scale
• Interpret the figure relative to attitude measures
for competing brands
• Multi Attribute Attitude Modeling (MAAM)
Multi-Attribute Attitude Models
n
Ab =  bi ei
i=1
Ab
=
attitude toward brand
bi
=
belief about the relationship
between brand and attribute i
ei
= attribute importance weight i
n
= number of salient attributes
Multi-Attribute Attitude Models
Example
Attribute
Value
(ei)
Store
X
Store
Y
Store
Z
Wide Selection
0.3
+2
+3
+3
Low Price
0.2
+3
-2
-1
High Quality
0.3
-1
+3
+1
Convenient
location
0.2
+2
+2
+3
biei for Store X: (0.3)(+2) + (+3)(0.2) + (-1)(0.3) + (+2)(0.2) = 1.3
biei for Store Y: (+3)(0.3) + (-2)(0.2) + (+3)(0.3) + (+2)(0.2) = 1.8
biei for Store Z: (+3)(0.3) + (-1)(0.2) + (+1)(0.3) + (+3)(0.2) = 1.6
Attitude Research
Attitude
Action/
Behavior
Three Components of Attitude
Cognitive
Component
Affective
Component
Action
Component
Attitude components
• Cognitive component
– Awareness of object
– Knowledge of attributes of object
– Judgments of
• importance of attributes of object
• Satisfaction
• Etc.
• Affective component
– Feelings and emotions
• Conative component
– ‘drive’ to act / behave – motivation
– Desire
• Attitude is a three – dimensional construct
Ideas, Concepts, Constructs and Variables
• E.g. “I want to make advertising that is “cool”,
“hip” and “edgy”
• Can you lay down clear boundaries between
“cool”, “hip” and “edgy”?
Ideas, Concepts, Constructs and Variables
• E.g. “I want to make advertising that is
“contemporary” and “effective”
• Can you lay down clear boundaries between
“contemporary” and “effective”?
Ideas, Concepts, Constructs and Variables
• 1. . E.g. How much do you like Winthrop
University?
• Hate it----Dislike it----neither like nor dislike ---Like it----Love it
– Variable:
• 2. E.g. “Did you buy the product when you last
went to the store?” Y/N
• Variable:
Construct vs. Variable
• Construct
– An idea / concept which
stands on its own
– In the conceptual /
abstract domain
– E.g. attitude,
satisfaction, love,
romance, commitment,
motivation, etc.
– May have several
dimensions e.g.
dimensions of attitude,
etc.
• Variable
– The operationalization
of the construct
– A variable can be
measured
– E.g. the
operationalization of
attitude is “liking”; of
romance could be
“attraction” etc.
– If a construct has
several dimensions, its
variable has several
factors e.g. factors of
attitude, etc.
Measurement and Scaling
• Measurement – standardized process of assigning
numbers / symbols to characteristics of objects
according to pre-specified rules
– One-to-one correspondence between the number /
symbol and the characteristic
– Assignment to be invariant over time and objects
• Scaling – process of creating a continuum on
which objects are located according to the
amount of the measured characteristic they
possess
Classification of attitude scales
Attitude Scales
Single-Item
Scales
Itemized
Category
Scales
Comparative
Scales
Rank-Order
Scales
Continuous
Scales
Multi-Item
Scales
Paired
Comparison
Scales
Constant
Sum
Scales
Pictorial
Scales
Semantic
Differential
Scales
Likert
Scales
Stapel
Scales
Continuous Scales
How would you rate Sears as a department
store?
Probably the worst -------------------------------------------- Probably
the best
Problems: Unreliable in interpretation hence not
widely used
Typical Attitude Rating Scales
• Single item scales – Only one item to measure
the construct
• Comparative
• Rank order
• Pictorial
• Constant sum
• Multi-item rating scales – More than one item to
measure the construct
• Likert
• Semantic Differential
• Stapel
Single Item rating scales
• Advantages
– Relatively quick, uncomplicated measurement
– Relatively simple to analyze
• Problems
– Can one item measure all the dimensions of the
construct?
Single item scales
• Itemized-category scales
– Labels each category on the scale
• Example:
• What is your overall satisfaction with McDonald’s
Hamburgers
–
–
–
–
Very satisfied
Quite satisfied
Somewhat satisfied
Not at all satisfied
What are the problems with this scale
Single item scales
• Comparative Scales – forces respondent to
evaluate the object w.r.t. another, on the same
attribute
• Example:
• Compared to Burger King, how would you rate
McDonald’s Hamburgers on taste
–
–
–
–
–
Very superior
Superior
Neither superior or inferior
Inferior
Very inferior
Single item scales
• Rank-order scales –
– requires respondents to arrange a set of objects with
regard to a common criterion e.g. interest in an ad,
brand preferences, etc.
• Closely corresponds with the choice process since
buyers make direct comparisons amongst
competing alternatives
Rank Order Scales
Please rank the following in order of your preference where
1 = your most preferred and 9 = your least preferred.
Brand
Brand
Brand
Brand
Brand
Brand
Brand
Brand
Brand
A
B
C
D
E
F
G
H
I
_____
_____
_____
_____
_____
_____
_____
_____
_____
What are the problems with this scale
• How will you improve this scale?
Single item scales
• Constant sum scaling
– Allocate a fixed number of rating points amongst
several objects / attributes to reflect relative
preference for the objects / importance of the
attributes
– Multi-attribute model importance weights
Constant Sum Scale
• Divide 100 points among the following attributes of a
PC in terms of how important they are to you in making
a purchase decision.
Clock Speed:
Hard drive size:
RAM size:
Price:
TOTAL
30
20
10
40
100
Possible problems with this scale?
Single item scales
• Pictorial Scales
– Various levels of the scale are depicted pictorially
– Generally used when surveying children / illiterate
samples
Pictorial Scales
•
Interviewer says: Eating Honey Munch Cereal
makes me feel:
Designing Scales
• Number of Scale Categories
• 2 to infinity (Problems?)
• 5 – 7 preferred
• Strength of the Anchors
• colorful vs. very colorful vs. extremely colorful
• Strong anchors are less likely to be used
• Balance of a Scale
• balanced vs. unbalanced (problems with unbalanced
scales?)
• Equal number of categories on both sides
Designing Scales
• Types of poles used in the scale
• Sweet and not-sweet vs. sweet and bitter
• Problems?
• Labeling of the Categories
• no labels vs. some labels vs. all labels
• Labeling reduces ambiguity
• Labeling also causes cracks
Designing scales
• Number of response alternatives
– Five to seven is a good number
– Two to three generally stifle responses and frustrate
respondents
– More than nine is superfluous
– An odd number is preferred since a neutral position
can be legitimately adopted
• “Don’t Know” option
– Use it when there is a distinct possibility
– Overuse may attract responses from disinterested
respondents
Multiple Item Scales
• Attitudes to complex objects like cars, insurance,
credit cards, etc. may have many facets
• Unrealistic to expect just one item to capture all
these facets
• Here we use multi-item scales
• Example: Attitudes to Winthrop University
Likert Scale
• Require respondents to indicate their degree of
agreement / disagreement with a variety of
statements related to the attribute or object
• Also called summated scales because scores on
individual items are summed to obtain scores for
respondents
Likert scale example – Satisfaction survey of Bank
Strongly Disagree Neutral
Disagree
Agree
Strongly
Agree
Courteous
service
1
2
3
4
5
Convenient
locations
1
2
3
4
5
Convenient
hours
1
2
3
4
5
Low
interest
loans
1
2
3
4
5
Semantic Differential Scale
• Used to describe a set of beliefs that comprise a
person’s image of an object
• Each scale item is bounded at each end by a
polar adjective or phrase / bipolar adjectives or
phrases
• Can be spatially represented on profile maps to a
clearer understanding
Semantic Differential Scale
Low Price
Consistent
Quality
Tangy
Bitter
1 1
High Price
Spotty
Quality
Smooth
Not Bitter
Stapel Scale
Heavy
+3
Consistent Quality
+3
Tangy
+3
+2
+2
+2
+1
+1
+1
-1
-1
-1
-2
-2
-2
-3
-3
-3
Exercise – Identify the scale
Exercise – Identify the scale
Accuracy of Attitude Measurements
• Reliability
– Does the scale perform consistently over time and over
different sets of respondents?
– Test-Retest reliability: administering the same scale at
two different points in time to the same / different
sample
– Absence of reliability induces random error in the
measurement
– Reliability of 0.7 and above is generally good
Accuracy of Attitude Measurements
• Validity
– Does the scale measure what it is intended to
measure?
– Absence of validity induces systematic error in the
measurement i.e. the scale is measuring something
else over and above the construct in question (e.g.
attitudes)
– A valid measure is one that reflects the true score
Accuracy of attitude measurements
• Observed score = true score + systematic error
+ random error
• Hence a valid measure has both zero systematic
and random errors
• If random error is zero (i.e. the scale is perfectly
reliable) it may still not be valid
– The scale may be consistently measuring something
else
• Hence reliability is a necessary but not sufficient
pre-condition of validity
Types of validity
• Face validity – a knowledgeable conclusion about
the scale validity
• Convergent validity
– Criterion validity – does the variable predict another
variable satisfactorily
• Does attitude to brand predict purchase intentions, both
measured at the same time?
– Predictive validity – if the DV is measured in the future
• Does college GPA predict the amount of salary you earn in the
future?
• Does attitude to brand predict future buying behavior?
Types of validity
• Discriminant validity
– Is your construct different from another construct
– Are attitude to brand and purchase intentions two
different constructs, or the same construct with two
different labels?
– Effect of attitude to brand and purchase intentions on
purchase behavior
• Construct validity
– Conclusion about the measure after testing reliability,
convergent and discriminant validity
Accuracy of Attitude Measurements
• Sensitivity
– Ability of the scale to capture meaningful differences in
attitudes
– Can be achieved by increasing the levels but the
greater the levels the lower the reliability
– Generally 5 to 7 levels are good
• Generalizability
– Ease of scale administration and interpretation in
different research settings
• Relevance
– Validity x Reliability (between 0 to 1)
– Meaningfulness to measure a construct
Accuracy of Attitude Measurements
• Dimensionality
– Does the construct consist of only one dimension or
more than one dimensions
– E.g. Attitudes – 1,2 or 3 dimensions?
– Measured through a factor analysis